Full disclosure: I worked for AMD for about a decade.
I agree with the bottom line here (we should spend gov't $$ to make sure significant amounts of chip manufacturing still get made in the US, and Intel has to be a big part of that). However, it misses the real reason for Intel's decline.
For most of it's existence, if Intel had missed a manufacturing node in Moore's Law, it would have been rightly recognized as an existential threat, and it would have become the #1 focus of the CEO and the rest of the company. Intel fell behind Moore's Law for years in its manufacturing, and it never caused them to appropriately panic until that hit the bottom line. This is a problem of an executive suite too focused on this quarter's results, and not focused enough on the engineering that drives those results. Previous generations of Intel executives understood that. It is likely that the current CEO, brought in once the board saw problems in their stock price, realizes it. If they had been run by executives who cared more about the underlying engineering, they would have had a 5-year head start in fixing the situation.
I think this is a result of the problem of financializaton. I don't think that an engineer has to run a company like Intel, Boeing or Apple. But if the growth of steady profits becomes a higher priority than what you actually produce, you are likely to milk the business at the expense of its long term prospects.
I mostly disagree with this line of reasoning. Yes, financialization is bad, but Intel would have been fine if its process went fine.
We probably will never know what exactly went wrong, but I think the current best guess is this: Intel set minimum metal pitch target for its 10 nm process at 36 nm, while TSMC set minimum metal pitch target for its 7 nm process (equivalent to Intel 10 nm process) at 40 nm. These processes were the last before EUV, pushing DUV to its limit. It turned out that you can do 40 nm but not 36 nm. That's it.
It was an extremely technical issue, with justifications on both sides, and Intel's decision wasn't obviously mistaken, even in retrospect. 36 nm was usual scaling to Moore's law. It was 40 nm that was unusual, TSMC underscaled metal relative to fin and gate. Why 40 nm? Because it was the limit of DUV double patterning. But Intel also knew that! That's why Intel went with quadruple patterning. It was a conservative choice to underscale metal to the limit of double patterning, but quadruple patterning wasn't crazy either. Everyone was using quadruple patterning for fin at this point, so it wasn't thought that risky.
Should Intel have reconsidered this once they encountered yield issues? Yes, but that's hindsight. All processes experience yield issues. Intel probably thought they can be solved, faster than going back to drawing board and doing everything again. Unfortunately they couldn't.
Yeah I generally agree with this take. Of course on HN and other technical sites the opinion "lol engineers should run companies" gets a lot of play, but plenty of engineers tank companies and don't get posted here with the alternative viewpoint.
Intel made a calculated bet, lost, and furthermore made a handful of other bad business decisions. We focus a lot on the latter part when in reality it's unlikely the "business" decisions were the primary reasons for failure... it very well may have been technical problems led by engineers in the first place.
I don't think Intel acted irrationally.
Last, this is the journey of almost all industries. The titans become too complacent, set for disruption by other smaller companies / outsiders. This is just the natural course of business due to inertia. Intel is hardly alone here.
EDIT: All that being said, the right man for the job is in the CEO chair. I personally believe he always should've had the role. We'll see how Gelsinger turns it around. I am optimistic.
> Yeah I generally agree with this take. Of course on HN and other technical sites the opinion "lol engineers should run companies" gets a lot of play, but plenty of engineers tank companies and don't get posted here with the alternative viewpoint.
Honestly, it's probably not so much that engineers should run companies, but that MBA/finance-types shouldn't (mainly because it seems their conventional wisdom has collected a lot of toxic ideas).
Engineers should be CEOs? Eh, probably not necessary and even a bad idea when you get past the early stage.
Engineers should run engineering companies? Yes, absolutely. The CEO is not the only position that "runs" the company - there should at the very least be a CTO that has equal political clout in terms of long term decision making.
You seem well versed on this so maybe you know: are there other strategic decisions that would have made the 40nm choice less risky? Often strategic management isnt about each bet being reasonable, it's about making a portfolio of bets that balance risks and give you a unique market position.
That's a better explanation for what went wrong at Intel. But I would also point to many other companies: Boeing, General Motors etc. that had really excellent products once upon a time and then faltered. Every case is very complicated and has many reasons.
A big part of this is investors demanding financializaton. Unless you're in a sector that investors have decided is a "growth" sector, it's all about capital efficiency and headcount and how much "fat" you can trim. See, for example, the recent gas price situation. Companies are far more worried about becoming overcapitalized than they are worried about losing out on some sales volume.
In general I don't believe that being an engineer is necessary to run a technology business (some of my best leaders were not), but I have come around to believing that Intel and Boeing are exceptions: the nature of their business requires new massive bets- not quite 'bet the company' but definitely 'bet the next 10 years of company performance' every so often (new fabs and new planes), and that one decision really needs to be made by an experienced engineer, due to the nature of the underlying evaluation. If not, then people who think they can bafflegab mother nature the same way they can auditors will make the decisions.
He started at Boeing in 1985 as an engineering intern, earned a masters in Aeronautics, and worked his way through management and engineering positions until he became CEO. I have no idea how much of the 737 Max issues could be placed on him directly, but he was CEO during its launch and the crashes.
The key 737 Max decisions were made circa 2011-2013, while James McNerney was the CEO (2005-2015). That was when the design objective of no simulator training required was set for the plane (and Boeing released marketing information promising that) while also promising to do it with engines that were too big for the wings. That decision was what led to everything else, and was loudly announced and marketed while Muilenburg was over on the Integrated Defense side of the company.
James McNerney was a Harvard MBA who spent time at McKinsey before going to the Jack Welch clown show at GE, where he ran the aircraft engine business for a long time. When he lost out to Immelt to be CEO there, he went to 3M for a few years then bounced to Boeing.
Now, Muilenburg has to bear the mistakes he made: pushing an unsafe plane through certification, not listening to the right people, not investigating the Lion Air crash sufficiently, etc. But the evidence I think strongly shows that someone who was not an engineer was making the final decisions on the "bet the company for the next 10 years" choice to go forward with the 737Max rather than a clean-sheet design, back in 2011.
It’s funny. I worked in a geography with a lot of GE alums. The ICs were really good for the most part, hard workers, smart but reserved.
The managers and executives turned consultants went out of their way to tell you about how they met Jack Welch or got screamed at by Jack Welch, etc. It was weird, and moreover, they almost without exception idiots.
Dennis Muilenberg also featured in the NOVA documentary "War of the X-Planes", around the middle of his career, where he appeared to be a risk-taking, forward-thinking engineering manager. Of note, I believe that a significant portion of his career was spent on the 'defense' side of Boeing, which some have blamed for the company's problems.
I personally place more blame on Muilenberg's predecessor, James McNerney, a former Proctor & Gamble, and 3M executive.
Just wanted to clarify that singling out Boeing and Intel specifically was not my intention: I really meant commercial aircraft and semiconductors industries as a whole, using those two companies as a syndoche, because of the massive investment and large time delays between the commitment of money and the result in the market. Those two in particular seem to require regular, hugely expensive (10+ billion dollar) efforts of 5-10 years in duration to keep up the pace of technical development, so picking the wrong horse is catastrophic for a long while.
(Obviously the A380 never killed anyone like the 737Max did, but it definitely was a massive, 25 billion dollar sinkhole for the company- some estimates are that Airbus never even turned a profit on the flyaway cost of the airplane, excluding development costs! If Boeing hadn't simultaneously had the 737Max actual disaster, it would have been much worse for Airbus. Intel has seen TSMC power ahead, which has made their problems more obvious.)
I think it boils down to the messaging that comes from top. Do they mainly talk about creating the best products or do they talk about shareholder value?
> For most of it's existence, if Intel had missed a manufacturing node in Moore's Law, it would have been rightly recognized as an existential threat, and it would have become the #1 focus of the CEO and the rest of the company.
Former Intel-er here. You hit it on the nail. Andy Grove used to say "Only paranoid survive" and acted accordingly. But when Paul Otellini became the CEO it was downhill ever since.
No, it was going downhill when Craig Barrett was CEO. He made lots of stupid mistakes, getting involved in too many risky and unnecessary ventures (like consumer products: remember the Intel keyboard and mouse? Or music player? Or worse, microscope?). The big one was tying the P4 processor to RAMBUS memory.
Yes, it started with Barrett but became painfully obvious under Otellini. When Steve Jobs asked Intel to build a chip for future iPhone Otellini said no -- one of the biggest strategic blunders. Also in 2005 Intel paid Dell not to use superior AMD offering instead of putting all hands on deck fixing their own (they did achieve parity eventually).
Intel with a bunch of Optane related presentations at #FlashMemorySummit
Some the Intel folks I spoke to about this (not pictured) had no clue it was going to be cancelled until they read it on an article from Toms Hardware.
Could it be opposite? Intel gave up Optane products so they release their research to community without worrying competitors catch up. No disclosure is due to pre-IR compliance.
They can't. It'd undermine the very existence of the ruling elite class that doesn't know how to do anything, besides 'managing' by following a magic formula taught at Harvard, Stanford, etc.
The problem is the existence of a business management layer as an idea - it's an attempt to get back to the good old days of feudalism, whose chief characteristics are nepotism, stagnation, corruption and nihilism, now that we've gotten rid of the mental pacifier of 'heaven' for the oppressed and hell for the ruling elites (religion).
this is the most resonant comment I've seen in months, as it entirely fits my experiences in engineering inside Apple, and later Amazon and Facebook.
the inner sanctum of very flat Apple was as far from feudalism as could be, for my several years, and it functioned like a very performant ecosystem of engineering. This diametrically contrasts precisely with Amazon engineering, where I saw all the horrors of feudalism and generalized nepotism (racism for example, sexism promoting incapable people to fit quotas, and flagrant corruption). Facebook was somewhere more like Apple but not in its DNA...just in the imitation is flattery sense.
Yes but I would argue Amazon has better software than apple by a long shot. Ignoring how you felt as a worker out what you read Amazon employees feel like, Amazon has better results.
It may be that Apple does more of the "inspiration" part of the work and Amazon does more of the "transpiration" part. There is software that is great on the outside but crap on the inside and vice versa. The last 80% of quality (after the first 80%) come from testing, discipline and other such such unsexy things.
As in can I justify my claim, or what are the reasons that Amazon produces better software? I would say the former is a qualitative statement and I am going to have a hard time backing it up with facts, and the latter I have no idea.
I run into non stop issues when interacting with apple software as an app producer, from Xcode to iOS to App Store to App Store connect. Their apis are pretty horrible, both the reliability and the design.
As an end user of apple products, I regularly run into odd software problems with my MacBook and iPhone. Every CarPlay experience I have had is buggy, both from the actual CarPlay connectivity as well as apples own CarPlay enabled apps.
Apple Watch is a horrible nightmare to make anything more complicated than a simple remote display.
I could go on and on. However on the Amazon side of things I have never experienced a significant bug, aside from the audible app freezing on occasion when interacting with CarPlay. I blame apple though based off how their own apps work. I say this as an extensive consumer of Amazon products both as a business owner, developer, and SRE, as well as an extensive user of Amazon consumer products.
All this being said I don’t like amazon as a business, I can find flaws in their decisions, but they are business decision flaws not just straight buggy products.
>I run into non stop issues when interacting with apple software...
>As an end user of apple products, I regularly run into odd software problems with my MacBook and iPhone. Every CarPlay experience I have had is buggy, both from the actual CarPlay connectivity as well as apples own CarPlay enabled apps.
>Apple Watch is a horrible nightmare to make anything more complicated than a simple remote display.
So why do you keep paying top dollar for Apple products? For work stuff, I can understand, but no one's forcing you to buy an Apple Watch.
Of course, many competitors to Apple products aren't perfect either, but they don't charge Apple prices either. I have some cheap Chinese fitness band and it doesn't work perfectly, but it was also only $35 so I really don't care as long as it mostly does what I want it to do, which it does.
I have to use an iPhone in order to test our highest grossing platform.
I’m not here to debate why apple has the most popular platform, I’m just stating what feels to me like a fairly unequivocal observation - apple’s software is pretty bad compared to Amazon’s. For whatever reason the company that takes millions in App Store revenue from my company, certainly doesn’t use it to make their software better!
>I have to use an iPhone in order to test our highest grossing platform.
As I said before: "So why do you keep paying top dollar for Apple products? For work stuff, I can understand, but no one's forcing you to buy an Apple Watch."
Your work making you use an iPhone for testing does not mean you're required to own and use an iPhone as your personal device, or buy into other parts of the Apple ecosystem. If you really dislike Apple's software so much, why are you still using it outside of work? You complain endlessly about Apple, but you happily continue to send them your money. Multiply this by countless other Apple consumers and you have your answer about why they take millions in App Store revenue from your company and don't bother to improve their software: why should they? If I ran a company and people shoveled their hard-earned money to me no matter how lousy my product was, why would I bother making my product better?
The nepotism is unreal in every company. Where I work the CEO hired a family member who makes more than me as a senior engineer and works 2-3 hours a day. Amazing.
The accountants took over the company, and the good engineers left.
I have witnessed this with other companies; and I am sorry they are never going to return to what they once were.
It is kind of obvious when you chat with all the sad employees: Why didn't you leave? And then you realise that the ones, who are left, are the ones who couldn't get a job elsewhere.
> The accountants took over the company, and the good engineers left.
As an accountant, I would quibble with this aspersion and lay the blame with management and other, broader issues.
> It is kind of obvious when you chat with all the sad employees: Why didn't you leave? And then you realise that the ones, who are left, are the ones who couldn't get a job elsewhere.
This seems true of many companies. Many of us are just playing musical chairs with our careers and jobs trying to find that job that doesn't suck.
I am sorry for insulting accountants and I realise what wrote is too harsh (but hey this is the internet so ...).
And you cannot have engineers running everything because then you will have a half-built Concorde and a bankrupt company.
But there is a balance. And if that balance tips for too long, you end up with the whole organisation content with just doing maintenance. And when that happens there is just no going back.
> I am sorry for insulting accountants and I realise what wrote is too harsh (but hey this is the internet so ...).
No worries. I didn't mean to sound angry. I appreciate the apology, didn't think it was necessary, and understand where you're coming from. I think you have valid points. I am very bad at reading and writing tone in quick comments online. I agree that having accountants and mba types in charge isn't a great idea most of the time or that accountants are necessarily the best people on average, but I am wee bit biased here against such blanket statements regarding my profession, as an accountant anyways.
In the aviation industry, Lockheed Martin famously avoided 'resting on its laurels' by having a separate 'skunkworks' part (who made the U-2, SR-71 and practically invented 'stealth' i.e. F-117, F-22, etc)
It was originally created by Kelly but he managed to transmit the spirit to a successor, who wrote an interesting book about his time (80's and 90's) called 'stealth'. If you look on Google you can find the PDF.
Another book about organisations having to deal with ossification is 'the soul of a new machine'. It's written by a journalist that was embedded in a R&D department of a computer manufacturer in the 70's and 80's. The place is run by someone a lot like Kelly Johnson, and the author managed very well to get into his mind. Again, I believe you can find a PDF on Google.
"It was originally created by Kelly but he managed to transmit the spirit to a successor, who wrote an interesting book about his time (80's and 90's) called 'stealth'."
The time period doesn't quite match up, but I found Skunk Works: A Personal Memoir of My Years at Lockheed by Ben R. Rich https://www.amazon.com/gp/product/0316743003/ to be quite interested if not quite as interesting as I'd hoped when I bought it.
Well, of course the blame is with management. Accountants shouldn't take over a company, and if they are leading the decisions, it's the management's fault.
It's quite like engineers taking over the marketing department.
I've noticed a pattern of how executives approach this:
1) When looking for more profits/revenue at a technology company, they'll often either try to push an R&D "big bet" to market half-baked, in hopes of it pulling in new revenue OR put the screws to product and engineering to churn out some "quick wins" to give them some momentum.
2) When focusing more on engineering, it's often because they're either distracted by other challenges faced by the organization outside of engineering OR they are up against an existential threat that mandates engineering investment and letting them steer the strategy somewhat.
I went to business school. The management textbooks used to tell you that steady profits were the most important thing. When I think about the things I was taught, particularly this incredible reverence for GE, I laugh.
"The management textbooks [at my business school] used to tell you that steady profits were the most important thing."
That's stark raving mad. It is for example one of the clearest signs someone is running a Ponzi scheme. Which would seem to transfer over to this not quite as dishonest domain.
Reverence for GE, a company doing technical things that started stack ranking and firing their employees.... Plus eventually got first their fingers and then last time I checked their whole body burned by going into financial businesses, very much not a core competence.
> That's stark raving mad. It is for example one of the clearest signs someone is running a Ponzi scheme.
Not really - I mean, if I run Toyota should I not be producing cars I can sell at a profit? And if my sales are falling, isn't that a bad thing?
Rising sales are good, of course - but not oscillating sales. If the rise is going to be followed by a sales slump in a few months time, do I want to take on extra staff who I'll then have to lay off, causing them untold strife and inconvenience? Do I want to build a new factory full of expensive machines, then have them stand idle while costing me interest and rent?
Industries like Software and IP licensing are, of course, often a different matter!
"Not really - I mean, if I run Toyota should I not be producing cars I can sell at a profit? And if my sales are falling, isn't that a bad thing?"
Yes to both. Maybe I'm misinterpreting the concept. "Steady" to me in this context means the same, thus financial shenanigans to make sure you don't report "excess" earnings in one period so you can have them in reserve for a future one where you have a slump.
"Regular" profits are ideal, but Toyota today is not a good example, it's a mature company in a country with a mature automobile industry. As a counter example, see the story sold to the investing community about Amazon, not that they would make a steady profit, but that they would become huge and then do something like that. Which they've delivered on (and I wouldn't judge them on anything in the COVID disruptive period 2020 to now).
I suppose the business courses you took are tech startup oriented. For tech companies until recently, the business is valued at roughly 10x revenue, so revenue is the all important metric. Profit is not important. This is why the barely profitable Salesforce and the non profit enterprise named Uber is worth so much in the stock market.
For traditional businesses, the metrics are profit and in particular profit margin (profit/revenue), so more revenue at same profit is bad. And other metrics like ROCE (Return on Capital Employed) and FCF (Free Cash Flow)... which incentivise lots of debt if you can borrow cheap but discourage capital investment. This is obviously antithesis of tech startups.
I guess the point is that you can make both cases. They would argue that a company with dominant sales and no profits could be bought by bought by a smaller competitor with profits.
But they had never seen VCs like today and could never conceive companies that could go for years with no profits and maintain high stock valuation. The world changes. So there's your answer: sometimes!
I read a lot of creative accounting case studies and am repeatedly surprised when others fall for the same simple tricks over and over again. At some point I have to assume willful ignorance.
I’m sure there are plenty of case studies in text books that management could read. Either they don’t read them or they take away the wrong message. Don’t forget the execs at these companies probably cashed in some big bonuses.
I heard this so many times. These goddamn CEOs are so myopic; all they care is next quarter's earnings. After the fall of every megacap, tech or otherwise (recently GE), someone always blames them for short sightedness. Is that really? Well, many CEOs of many huge corporations just plan for the next quarter, and they seem to be doing just fine.
The average large corporation is typically in a stable/stagnant industry that is nowhere near as competitive as chips, where the stakes are 100x higher.
They get by being the largest kid on the block and few young companies chomping at the bit, and a series of B/C-tier competitors even dumber and poorly run that they can't even do the next quarter effectively.
Countless mega-corps are just floating by with the ebbs and flows of a market, burning off some early moment of success. Staying at the top longer than a decade or two (usually the timeframe where other market entrants take notice of your weakness) is the real test.
Otherwise just being a decent manager/leader that keeps the existing established business model chugging along and hiring slightly above average executives to put out fires is good enough.
Well not all CEOs are so myopic; Andy Grove of Intel was not, for example.
Whether or not a next-quarter focus works, probably depends on the industry. In semiconductors, especially high-performance semiconductors, the R&D on manufacturing this year determines whether the design several years from now is able to be competitive, so there is a need to be more forward looking.
You could apply this with slightly different specifics to many US national and multinational corporations.
The efficient market rewards inefficient stock buybacks, record debt fueled dividends, years of underinvestment, low wages, and most of all no strings attached bailouts. They may be a loser, but we'll pay for their loss. Socialize the losses/costs and privatize the gains.
Based on the performance of publicly listed companies in the last 15 years, the stock market has rewarded companies with great cash flow, high profits, and relatively rosy outlooks.
Of course there is an underlying assumption the government will bail out simultaneous/systemic big failures, but it does not seem accurate to assume any random business will be bailed out.
> but it does it [sic] seem accurate to assume any random business will be bailed out.
Precisely the issue. It's not "any random business". It's any business with the appropriate properties and connections. If it was truly just random business, perhaps the bail outs would not create such a level of moral hazard. But it's not, and businesses know that.
I don't like the bailout but I understand they were necessary. I just wish they came with harsher deterrents. Like if the shareholders vote to take a bailout, the Board, C-Suite, and VP's all get let go and any financial packages can be litigated in special bailout courts.
>The reason this matters is because chips matter for many use cases outside of PCs and servers — Intel’s focus — which is to say that TSMC matters. Nearly every piece of equipment these days, military or otherwise, has a processor inside. Some of these don’t require particularly high performance, and can be manufactured by fabs built years ago all over the U.S. and across the world; others, though, require the most advanced processes, which means they must be manufactured in Taiwan by TSMC.
>This is a big problem if you are a U.S. military planner. Your job is not to figure out if there will ever be a war between the U.S. and China, but to plan for an eventuality you hope never occurs. And in that planning the fact that TSMC’s foundries — and Samsung’s — are within easy reach of Chinese missiles is a major issue.
Regarding your second question, what companies and countries do you have in mind?
For the second question I meant: if everyone is trying to build chips at home, at best we get an expensive and inefficient arms-race for subsidies. At worst we get trade-wars or actual wars.
I think that rationale would apply to some other countries, but not all. For example, Japan should want to maintain a viable semiconductor manufacturing capacity (and might be at risk, currently). Now obviously Luxembourg cannot. Where exactly in between the size of Japan and Luxembourg one draws the line is a tough question, but probably western Europe has at least a couple countries (Germany, France, UK) who are big enough.
Lower-income countries like Brazil, South Africa, Argentina, India probably want to, but may have to work their way up to that; it's an expensive business to be in.
I notice that China has absolutely made it a top priority to try to get into this, despite having some difficulty doing so.
Because you can't make war without chips. During war, global supply chains are cut off. And if you can't make war, then you can't have a global order on your terms. Instead, you have a global order on the terms of the Chinese. They become powerful and rich. You (and your allies) become weak and poor.
You make chips in the US for the same reason you make nukes in the US.
It applies to other countries only insofar as they want to be independent. Most Western countries can shelter under the US umbrella. Again, same as with nukes.
I'm usually a big fan of stratechery, but for once I disagree with the explanation of Intel's decline.
I think it's much simpler: being number one in fab technology is mostly driven by volume: whoever has the most volume wins. During the PC era, Intel had the most volume, so they also had the best fabs. It's a winner-take-all, self-reinforcing ecosystem.
But in the last 10 years, mobile has had the highest volume. That's what allows TSMC to have the best fabs, and Intel is struggling. That's all there is to it. Nothing to do with services or software.
This would the correct explanation if Intel would have made no mistakes.
However, it is beyond reasonable doubt that Intel has made some extremely ugly management mistakes during their failed attempts to transition from 14 nm to 10 nm, which have absolutely nothing to do with the competition from TSMC or others.
Maybe Intel would have lost their technological advance regardless how the company had been managed, due to the difference in volume between smartphone chips and PC chips, but now we cannot know which has mattered more, the internal mismanagement or the change of the semiconductor device market favoring their competition.
I wonder if it will ever become public knowledge which were exactly the technology problems of Intel and who are those guilty for the decline at the company, because the failures could not have had any real technical causes but only bad management causes.
During their worst years Intel has been capable even of presenting to the public fake technical presentations about the alleged great characteristics of their future 10 nm CMOS process, which is something that I would never have expected from a company like Intel. The marketing messages from any company are expected to be full of lies, but the technical presentations are expected to match reality.
During many years, unlike TSMC, Intel has behaved as if it were completely unable to predict the precise characteristics of the CMOS process that they will able to manufacture in the following year.
How could this happen is very hard to understand, as the manufacturing processes and the future devices made with them can be simulated long in advance of their implementation, and the models can be tuned continuously while the processes evolve, by fabricating and measuring various test structures. Unlike for Intel, for TSMC the performance predictions for future processes have always been reasonably accurate.
> I wonder if it will ever become public knowledge which were exactly the technology problems of Intel and who are those guilty for the decline at the company, because the failures could not have had any real technical causes but only bad management causes.
Why do you think this? It was a yield issue, and the best guess is it was caused by multi patterning. (Intel themselves said so.)
The yields are indeed hard to predict, so if that would have been the only problem, then there would not have been so much reasons to blame the Intel management.
As I have already mentioned, Intel has behaved as if being completely unable of predicting the electrical characteristics of the devices made with their future 10-nm processes.
The first generation of 10-nm products was Cannon Lake, launched in 2018. I have actually bought out of curiosity the Intel NUC with Cannon Lake U, together with an Intel NUC with Coffee Lake U, a 14-nm CPU launched simultaneously with Cannon Lake U.
The 10-nm CPU had pathetic performances in comparison with the 14-nm CPU, i.e. much lower clock frequency at identical power consumption or much higher power consumption at identical performance.
This too low performance should have been determined very early during the design of Cannon Lake and the project should have been canceled even before reaching the tape-out phase.
Even the 2nd variant of the 10-nm process, used for Ice Lake in 2019 was inferior to the by then 5-year old 14-nm process. This should also have been known enough in advance to cancel Ice Lake before tape-out.
Converting any Intel factory to mass production of 10-nm wafers should have never begun before having a working 10-nm process able to make electrically-better transistors than in the old 14-nm process.
Intel has begun the mass-production of chips using the 14-nm process in 2014. By then they must have already started the design of the 10-nm process, but only in 2020 they have succeeded to make better 10-nm transistors than their 14-nm transistors (while still having a lower maximum clock frequency) and only in 2021 they have succeeded to exceed the old transistors in all parameters.
It is impossible to understand why have they wasted huge amounts of money for failed projects like Cannon Lake and Ice Lake and for factories using uncompetitive process variants, instead of concentrating all their resources to their top priority of finding a way to manufacture 10-nm transistors that are better than their existing 14-nm transistors, which is normally a precondition for starting any design process of a CPU using a new process and for building a new factory.
It all looks like if most people working at Intel would not have been aware that they actually do not have any working 10 nm manufacturing process, and they were continuing all their usual activities, like designing CPUs, but with fake transistor models, or building new factories, while there was some gang who knew that the 10-nm process does not exist outside fake presentations, but they hoped that there will be some miracle and the 10-nm process will just work when the time comes to use it.
Intel has a decades long history of very poor top level engineering management. They got petrified about the issue of how much DRAM they'd be able to put in a system in the 1990 (remember Rambus RDRAM?), and among the consequences was a 1 million part recall of motherboards shipped to Dell just before they were going to be shipped to customers, plus another related one I can't remember the details of yet.
This speaks to the sorts of dysfunction you're speaking of; certainly someone at Intel realized these parts weren't going to work in the real world.
The CISC vs RISC point Ben makes is key here. Intel could have been a big player in mobile: it put too much focus on x86 though when it could have been making (its own or others) Arm based SoCs.
Edit: not that Arm is necessarily RISC - but it certainly takes an approach that is different to x86 and better suited to mobile.
I've lost count how many misguided bids of taking x86 to low-power they made. Also the apparent unshakeable confidence that ARM will never become viable on server and desktop.
In this context the Gelsinger video clip in the article is fascinating. Sometimes companies fixate on one element of their strategy that has worked in the past.
It seems that Intel had ‘always make x86 / backwards compatible’ as their fixation.
Intel tried to make Arm SoCs. But their customer management sucked, so in the end they only manufactured the SoCs that they designed, and didn't sell enough of them to be sustainable.
Also perhaps there's a role where software/architecture just got better - they didn't have LLVM or other forms of code translation (if I'm phrasing that right) back then which seems to make the cpu architecture less of a critical issue.
That is one of several reasons proposed in TFA; in particular TFA mentions TSMC's earlier investment in EUV being due to the guaranteed volume of smartphone chips.
Did not read TFA and if it doesn't focus a lot on the 10 nm/Intel 7 botch it's not fine, but I've read TSMC's very roughly equivalent 7 nm node was also non-EUV but less aggressive than Intel's. And it succeeded unlike Intel, which then made it natural to start introducing in production EUV in their 7+ node. Here I suspect gaining real experience in chips delivered to customers trumps whatever volume you'll end up doing.
I very much doubt this. When EUV decision was made Intel was at the top of the world and had more volume than TSMC. As I understand Intel's reluctance was mostly about doubt whether ASML could deliver, and while ASML delivered I must say, even in retrospect, Intel's doubt was justified.
Intel is in big trouble and the only card they have left is the nationalism card. And that card is for losers.
We are about to enter a silicon slump with low demand and oversupply; for Intel things will get worse.
What will happen when it is clear to everyone including politicians that all that subsidy money is going into a black hole? And people finally figure out that a rocket that blows up stuff doesn't need a 2nm chip?
They are going to lose interest and those subsidies that is Intel's last hope will end.
"We are about to enter a silicon slump with low demand and oversupply"
What causes you to assume that?
"What will happen [if...] people finally figure out that a rocket that blows up stuff doesn't need a 2nm chip?"
One, they'll learn that we're not doing too hot on custom chip fabs in general? They'll also, if they're really that dense, learn that we need chips for a few more things than to make things go boom. (FWIW, they're not that dense.)
"They are going to lose interest and those subsidies that is Intel's last hope will end."
Even if the rest of your statements were correct (and I don't think they are), that is absolutely not how subsidies work. Local subsidies translate to local jobs translate to votes. Keeping them running is rather important to politicians. So, no, "lose interest" is not exactly the most likely failure mode here.
Those 2nm 5nm 7nm shit has been driven by people's need for new mobile phones, and people are getting tired of buying new phone every year without much persuading new features.
Also cryptocurrency mining has lost its momentum, lots of networks will switch to PoS to replace the CPU/GPU hungry hashing.
Companies have been filling their stockpiling, and anticipating years of demand so they have a safety buffer for many months already. It's natural that this is followed by low demand.
At the same time, the world seems to be entering a recession.
I've read that the problems with chips for cars are actually old processes (greater than 100nm) that are commodity and so no one really cares to make them and the equipment has one foot in the grave and such, so I don't think anyone cares too much about putting chips into rockets since that stuff was figured out awhile ago.
My perception of national defense types is somewhat influenced by reading their press stuff and somewhat influenced by General Buck Turgidson from Dr Strangelove. He would say, "We must not have a $weapon gap!" (missiles, mineshafts, whatever he perceived to be in the national interest at the moment). And today, what occupies the minds of these types is AI which means latest gen chips and problems with pesky nerds and their ethical considerations. They greatly fear that China has surpassed the US in AI and has us outgunned. To be fair to them, AI could certainly be a potent weapon, but I have no idea how their perceptions of China reflect reality.
> "The most important decision was shifting to extreme ultraviolet lithography at a time when Intel thought it was much too expensive and difficult to implement; TSMC, backed by Apple’s commitment to buy the best chips it could make, committed to EUV in 2014, and delivered the first EUV-derived chips in 2019 for the iPhone. Those EUV machines are made by one company — ASML. They’re worth more than Intel too (and Intel is a customer)."
CNBC toured ASML clean rooms in a very interesting video. IIRC, the only thing keeping ASML from exporting these machines to China is some EU/USA export control rule:
Great piece but it seems that Ben doesn't quite finish joining the dots.
If Intel is to have a successful foundry business at the cutting edge then its customers will be those firms that are designing cutting edge CPUs, SoCs and GPUs.
That means Apple & Qualcomm (possibly OK) but also Nvidia, AMD and other Arm based competitors with Intel's CPUs and GPUs.
Can't see how this will work with Intel in its current form.
That’s a fair challenge. I think it’s a little different though - the SoC was one component of the iPhone - so full details of the competing product weren’t being shared.
Plus when the relationship started volumes were small for both iPhone and Samsung phones. As we know Apple ended the relationship eventually.
I struggle to see Lisa Su handing this info to Intel even with some strong safeguards.
> The most important decision was shifting to extreme ultraviolet lithography at a time when Intel thought it was much too expensive and difficult to implement; TSMC, backed by Apple’s commitment to buy the best chips it could make, committed to EUV in 2014, and delivered the first EUV-derived chips in 2019 for the iPhone.
Can somebody here correct me if I am wrong, but my impression was that Intel did commit early to EUV, with initial plans to start high volume manufacturing in line with other fabs (initial schedule was to introduce EUV in 2017 [1], it just got postponed many times), but they just failed in their execution.
I think your citation is 9 years old. What I've always heard is what they called 10 nm was non-EUV. You are correct that what they called 7 nm was going to use EUV, pretty much had to, but 10 nm catastrophically failed and the company responded very poorly to that.
Not that long ago I remember reading, forget at what confidence level, that their old 7 mn was believed to have the potential to leapfrog their 10 nm and save the company, but others said they had some problematic things in common (not counting the very poorly run company!) that made that unlikely. In any event what was 7 nm did not ride to the company's rescue.
Their 7 nm process has been renamed now as "Intel 4" and there is a presentation of it at Anandtech, which looks more credible than the fake presentations of the 10 nm process of some years ago.
The first product using this process, Meteor Lake, is expected 1 year from now.
Meteor Lake/Intel 4 would then be six years late if the wccftech.com article's commentary on the two nodes was correct. Good to hear about the presentation material, thanks!
Chips are great, but you need discrete components, circuit boards, and other support, and the manufacturing infrastructure for all of that. Do we really have all of that in sufficient quantity domestically any more?
Intel tried very hard to get Apple as a customer for the iPhone. They knew very well of mr. Christensen but they simply could not deliver hinting that Intel's problems are not that of strategy but at a lower more fundamental level of competence and ability to organise.
(The stuff about RISC/CISC is a red herring - it basically doesn't matter on a modern cpu.)
Here is an article about Intel and Christensen with clip of him from 2012 (or earlier) where he mentions his work with Intel.
I think Intel "tried very hard" from the point of view of a small company, but the fraction of Intel's resources put behind it was minuscule and felt very perfunctory.
Right or wrong, they had zero interest in betting the future of the company on low-power chips for mobile, or any other market.
[edit]
I just read the wikipedia summary of the prescription for incumbents in The Innovator's Dillema and the one thing Intel was missing was:
> They allow the disruption organization to utilize all of the company's resources when needed but are careful to make sure the processes and values were not those of the company.
To a certain extent, it feels like buying a cloud offering from AWS vs. Google. Google might have a lot of resources dedicated in absolute terms to the cloud offering, but it feels like an afterthought that might go away at any moment. The AWS offering feels like a real product that Amazon plans on making money from.
In both the cases of Intel and Google, I can't say which is true, but customers can only make their plans based on appearances, not reality.
> This is why Intel’s shift to being not simply an integrated device manufacturer but also a foundry is important: yes, it’s the right thing to do for Intel’s business, but it’s also good for the West if Intel can pull it off.
AMDs spin-off of Global Foundries only gets mentioned once as an aside, yet surely there are multiple important reasons why that story is relevant to Intel.
I am a Libertarian so this may seem like an odd comment: I support some government intervention to design and manufacture of at least low tech chips (adequate for laptops, tractors, industrial uses in general).
I am not a hardware expert, but given that the Open RISC projects seem very worthy of support, both in design efforts and multiple inside the US fab facilities.
TSMC’s new 5nm factory just opened 60 miles south of where I live in Arizona is also the kind of progress I want to see. I am a fan of globalization but we need to balance that with good “Plan B’s” for building what we really need inside the US.
US is not lacking in CPU design. I agree OpenRISC (and RISC-V) is worthy of support, but they benefit China more than US, and it would be poor use of US tax money to benefit China more than US.
I'm a 'government using funding to improve the world' vs 'govt using funding to control people' libertarian.
Scientific r and d falls strongly into the former category.
DEA, FBI, NSA, CIA, TSA, ATF, DoD, state police, county police, municipal police, neighborhood police, and other organizations used to control people. Plus the literally uncountable number of criminal and regulatory laws on the books that keep growing exponentially without congress taking time to go back and review old laws...
May be similar along my lines. I am for doing something for the silicon industry, I am just not into dumping into established giants. I’d love to see us throw money at people with novel ideas, new and interesting designs not though of before. The problem is, that is playing the long game when all we care about is winning the short game.
Yeah my take is not that he's hypocritical but that like most people he's not an actual libertarian. To be libertarian is to know there are things that government intervention can do that will benefit you but still know to forgo those things because their costs are higher to society as a whole. What he is is just a republican.
That is how I see it. I often drift into hypocrisy: I am 100% anti-war except if my country was directly attacked, but in the financial interests of me and my family, I do own “defense” stocks.
Also in my own self interest: I am for our government massively supporting poor kids with meals at schools, special tutoring, etc. in order for us to have a larger and better educated work force in the future.
fwiw it didn't come through as a rant. Most of us if we are honest with ourselves have internal contradictions. It's important to be aware.
My wife is fundamentally a socialist (we are in Canada so it's not a misunderstood bad scary word :-). But she is also a home depot manager, where she is the voice of capitalist market forces. What she wishes the system should be, and the system she is daily a part of, are frequently at odds.
I personally find extremes of any given philosophy to be impractical and untenable. I respect more people who, like it appears yourself, say "this is my general leaning,and these are my pragmatical exceptions".
In today's world you can't really follow any ideology in a clean way. If you want to live you have to make compromises somewhere and basically be a hypocrite. I only have a problem with people that don't admit that (of which there are many and they are very loud).
>"the goal should be to counteract the fundamental forces pushing manufacturing to geopolitically risky regions, and Intel is the only real conduit available to do that."
I wonder if it's fair to assume based on this the US gov will prop up intel as a foundry for as long as necessary, to secure it's own supply and reduce reliance on Taiwan/TSMC.
I wonder to what extent Washington's fuss over Taiwan is simply TSMC lobbying at work. They're an important company that is gonna get their lunch eaten by the mainland. But what can Washington even do about it?
Beyond the "unsinkable aircraft carrier" angle, that is.
"[TSMC is] an important company that is gonna get their lunch eaten by the mainland."
How so? Not for the foreseeable future in bleeding and leading edge nodes, where the PRC is some unknown but very long distance away from making their own EUV machines.
An invasion of Taiwan just means an end to TSMC production there, a point its chairman Mark Liu made a few days ago in a CNN interview. That would have really bad effects on the PRC's narrow economy with its export focus.
LIU: Nobody can control TSMC by force. If you take a military force or invasion you will render TSMC's factory non-operable because this is such a sophisticated manufacturing facility. It depends on the real time [correlation] with the outside world, with Europe, with Japan, with the U.S., from materials to chemicals to spare parts to engineering software diagnosis. And it's everybody's effort to maybe this factory operable. So if you take it over by force you can no longer make it operable.
China doesn't need to take TSMC by force. All they need to do by force is install a puppet government in Taipei.
No bombs will fall anywhere near TSMC's facilities. No tanks will go anywhere near their employee parking lots. Life will go on just like before. For a few years, at least.
It'll be just like Hong Kong. No bloodshed there either.
Come on. There is not a less "narrow" major economy on the planet than China's. This is just silly.
USA would collapse in weeks if China blocked all exports. China would run out of dollars, which coincidentally would become worthless at the same time.
While I don't accept the way you've reframed the decision criteria, I can point out one US nuclear attack sub busting a few ships loaded with fuel and food for the PRC "would collapse in weeks." Which is relevant since we're discussing my claim of how "narrow" the PRC's economy is, not anything about the US.
> Which is relevant since we're discussing my claim of how "narrow" the PRC's economy is, not anything about the US.
The US economy is deeply dependent on the Chinese economy. It is a total fantasy to pretend that the Chinese economy can suffer without US economy suffering massively as well.
Again, what's your point in trying to change the topic of the discussion? You certain Xi is going to further trash his own economy merely to further hurt the US economy?
For that matter, I think your original proposition has been partly proven wrong, the traffic jams at West Coast ports have deprived the US of timely container deliveries from the PRC and Asia in general. Some high value low weight and volume stuff got through via air, but somehow we continued to operate a modern technological industrial civilization.
The US has been sending carrier groups to support Taiwan long before they were a leading edge semiconductor manufacturer. So it's geopolitics as normal.
As for getting crushed by SMIC, this is only with possible massive government investment from the PRC. Now you see many other governments reacting with subsidies in kind. It will be a totally distorted market in 10 years time.
Hence "to what extent". TFA implies that TSMC leaders were VIPs at the Pelosi meeting. What does this have to do with Washington's long-standing policy of militarizing Taiwan?
Chinese semiconductor companies. If this insanity continues (nationalism and sanctions), the massive Chinese market will end up being serviced only by Chinese chip companies. Some of these are investable via the Chinese or Hong Kong stock markets.
I agree with the bottom line here (we should spend gov't $$ to make sure significant amounts of chip manufacturing still get made in the US, and Intel has to be a big part of that). However, it misses the real reason for Intel's decline.
For most of it's existence, if Intel had missed a manufacturing node in Moore's Law, it would have been rightly recognized as an existential threat, and it would have become the #1 focus of the CEO and the rest of the company. Intel fell behind Moore's Law for years in its manufacturing, and it never caused them to appropriately panic until that hit the bottom line. This is a problem of an executive suite too focused on this quarter's results, and not focused enough on the engineering that drives those results. Previous generations of Intel executives understood that. It is likely that the current CEO, brought in once the board saw problems in their stock price, realizes it. If they had been run by executives who cared more about the underlying engineering, they would have had a 5-year head start in fixing the situation.